Data-Driven Prediction of Compressive Strength in POFA-Modified Concrete: A Comparative Study of LR, NLR, and ANN Models

Authors

• Shuaaib A. Mohammed, Civil Engineering Department, Collage of Engineering, University of Sulaimani, Sulaimaniyah, Kurdistan Region, Iraq
• Hawdin Ismael Ibrahim, Civil Engineering Department, Collage of Engineering, University of Sulaimani, Sulaimaniyah, Kurdistan Region, Iraq
• Diar Fatah Abdulrahman Askar, City Planning Engineering Department, Technical College of Engineering, Sulaimani Polytechnic University, Sulaimani, Kurdistan Region, Iraq.
• Hersh F Mahmood, Civil Engineering Department, University of Halabja, Halabja, Kurdistan Region, Iraq
• Mehdi Khodaei, Faculty of Materials Science and Engineering, K. N. Toosi University of Technology, Tehran, Iran.
• Soran Abdrahman Ahmad, Civil Engineering Department, Collage of Engineering, University of Sulaimani, Sulaimaniyah, Kurdistan Region, IraqFollow

Subject Area

Civil and Environmental Engineering

Article Type

Original Study

Abstract

Production of concrete results in high CO₂ emissions, high energy demand, and consumption of various raw materials; however, it is one of the most widely used materials in construction to date. Therefore, researchers aim to minimize the impact of concrete production through various methods, such as introducing sustainable sources of materials that can be used in concrete. For instance, agricultural waste materials can be used as partial cement replacement. This study presents a comprehensive data-driven analysis of the effects of POFA on the properties of normal-strength concrete. By synthesizing a database of 196 experimental datasets from existing literature, three predictive models, Linear Regression (LR), Non-Linear Regression (NLR), and Artificial Neural Networks (ANN), were developed to forecast compressive strength. To determine the most effective model in this study, various statistical parameters were used, such as Coefficient of Determination (R²), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Scatter Index (SI). The analysis indicates that while POFA-modified concrete exhibits reduced slump and heat of hydration alongside extended setting times, an optimal replacement level of approximately 20% maximizes long-term mechanical performance. Ultimately, the ANN model demonstrated superior predictive accuracy over regression-based techniques, yielding the highest R2 and lowest error values, thereby highlighting the efficacy of machine learning in interpreting complex historical data to optimize sustainable concrete design.

Keywords

Compressive strength, POFA, Modeling, ANN, Normal Concrete

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 License.

Recommended Citation

Mohammed, Shuaaib A.; Ibrahim, Hawdin Ismael; Askar, Diar Fatah Abdulrahman; Mahmood, Hersh F; Khodaei, Mehdi; and Ahmad, Soran Abdrahman (2026) "Data-Driven Prediction of Compressive Strength in POFA-Modified Concrete: A Comparative Study of LR, NLR, and ANN Models," Mansoura Engineering Journal: Vol. 51 : Iss. 6 , Article 1.
Available at: https://doi.org/10.58491/2735-4202.3473